Hi Gabriel,

Thanks — that is a very useful observation.

I agree that, at a surface level, the SD-JWT presentation syntax:

<Issuer-signed JWT>~<KB-JWT>

captures an important idea: combining a credential with a
proof-of-possession artifact in a compact presentation form.

That said, I think there are a few reasons why directly reusing the SD-JWT
syntax outside the SD-JWT presentation model may not map cleanly to OAuth
sender-constrained access token use cases.

In SD-JWT, the KB-JWT is fundamentally a detached presentation proof:

   -

   it commits to a disclosure set via sd_hash
   -

   it binds to a verifier via aud
   -

   and it depends on a verifier-provided nonce

The ~ serialization is therefore tightly coupled to the SD-JWT
disclosure/presentation processing model.

In contrast, in the OAuth sender-constraining case, the requirements are
somewhat different:

   -

   the proof often needs to bind to request context (resource + method)
   -

   deployments may require offline proof generation (no verifier nonce
   round-trip)
   -

   and the credential/proof pair ideally survives intermediary forwarding
   as a single inseparable object

One consequence is that EPOP treats the proof and credential as a single
signed envelope, rather than two correlated artifacts linked externally by
presentation syntax. The nested-token (ntk) approach was intended to
preserve standard JWT processing semantics while avoiding special parsing
logic around ~-delimited compound objects.

So I think the two models are conceptually related, but optimized for
different layers:

   -

   SD-JWT KB: holder-binding for selective disclosure credential
   presentation
   -

   EPOP: sender-constraining for OAuth/API authorization across transports

That said, I do think there is value in aligning terminology and conceptual
models where possible, and your point about reuse is well taken.

Thanks again for raising this.

I have captured detailed analysis of this draft with RFC 9901 here:
https://github.com/asambeka/epop/blob/main/AnalysisWithRFC9901.md

Best,
Ashwin

On Sun, May 17, 2026 at 1:26 AM Gabriel Corona <[email protected]>
wrote:

> Le 17/05/2026 à 04:08, Ashwin Ambekar a écrit :
> > EPOP defines a single unified token type that carries both the
> > credential (authorization code, access token, or refresh token) and its
> > cryptographic proof of possessio
>
> For what its worth, SD-JWT [1] already has a mechanism for associating a
> key binding proof of possession to a JWT in a single token:
>
> <Issuer-signed JWT>~<KB-JWT>
>
> Would it make sense, to reuse this syntax in other contexts?
>
> [1] https://datatracker.ietf.org/doc/html/rfc9901
>
> Regards,
>
> Gabriel
>
> _______________________________________________
> OAuth mailing list -- [email protected]
> To unsubscribe send an email to [email protected]
>


-- 
Regards
Ashwin Ambekar
_______________________________________________
OAuth mailing list -- [email protected]
To unsubscribe send an email to [email protected]

Reply via email to